Full review
Fetch My Epic Token CAIHL draft report
Evidence-linked HugoScore draft report for a health AI tool that affects patients.
Agency checklist · October 30, 2026
Expands 10 of 12 yes
Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.
Assessed function: Patient-directed record retrieval and AI-enabling infrastructure. The core token utility does not itself provide AI clinical judgments.
A no can mean missing documentation or institution-dependent control; it does not by itself establish harmful behavior. These categories are not clinical safety or product-quality ratings.
- Public documents do not establish actual product behavior.
- The independent review screened rules and selected evidence; it did not replicate all twelve answers for this tool.
- Host Q12 and the AI/core-utility boundary remain unresolved; this Expands placement is not independently confirmed.
- October 30 follow-up: the non-AI retrieval utility and the suggested Gemini workflow need separate, consistent scope treatment. No answer or category changed. A 9/12 calculation from changing Q12 alone is hypothetical, not an accepted reassessment.
See all twelve answers and their evidence
| Question | Answer | Evidence / review | Source and reasoning |
|---|---|---|---|
| 1. Told it is AI | Yes | documented Not independently replicated | Using today's AI platforms you can build your own custom MyChart app without having any software background at all!Source Scope is the AI-enabling pathway: the HowTo page names Google Gemini and tells the user to paste a prompt into it. The core token and ZIP utility contains no AI. |
| 2. Operator and model maker named | No | not-documented Not independently replicated | Quotation not recorded; evidence repair pending. The operator appears only as the GitHub handle glmck13 ('Copyright (c) 2023 glmck13' in the MIT license, 'my Github site' on the pages). No real name or organization is published. Google is named as the AI maker for the suggested Gemini step. |
| 3. Patient chooses to use it | Yes | documented Not independently replicated | you can access your health records directly without going through MyChartSource The patient chooses to log in and fetch their own token and data. |
| 4. Patient sets the task | Yes | documented Not independently replicated | Try modifying the prompt to download your CBC/CMP results to a CSV file, or retrieve your immunization record.Source The patient writes or edits the prompt and decides what to pull and analyze. |
| 5. Patient can decline or stop | Yes | documented Not independently replicated | you can access your health records directly without going through MyChartSource Self-started tool rule: Q3 is yes and no care depends on it. Source is the Q3 source. |
| 6. Patient sees the AI output | Yes | documented Not independently replicated | saves this as a table within an HTML file on diskSource The patient runs the generated script and sees its output directly. The AI-generated code is visible before it is run. |
| 7. Patient can correct or challenge | No | not-documented Not independently replicated | Quotation not recorded; evidence repair pending. No documented route to correct or contest AI output. The Feedback page only points to GitHub for site problems. The patient can re-prompt, but that is not a documented challenge route. |
| 8. Patient can get a copy | Yes | documented Not independently replicated | packages these in a ZIP file for downloadSource The Download EHR button returns a ZIP of visits, labs, reports and notes. The same tokens let the patient pull FHIR data with any script. |
| 9. Patient can delete | Yes | documented Not independently replicated | does not collect, retain, or distribute any personal information from the end userSource Nothing is held to delete. The downloader script works in a temporary directory with 'guaranteed disk cleanup' and streams the ZIP to the user. This is no-retention, not a deletion feature. |
| 10. Training or sale does not override patient choice | Yes | documented Not independently replicated | does not collect, retain, or distribute any personal information from the end userSource No user data is kept, so none can be used for training or sold. Applies to the site. If the patient pastes results into an AI host, that host's terms apply (see caveats). |
| 11. Patient can run it or take it elsewhere | Yes | documented Not independently replicated | made available free of charge under the terms and conditions specified by the MIT licenseSource MIT-licensed open source, code on GitHub. The token site is also reproducible by the patient. Note: running your own copy needs the patient to arrange Epic app registration, which I did not verify. |
| 12. No advertising or data business | Yes | documented Not independently replicated | does not collect, retain, or distribute any personal information from the end userSource Policy also says no cookies are generated or used. No ads, sale or licensing documented. Host caveat applies if Gemini is used for analysis. |
Initial coding: Claude Sonnet helpers, compiled by Claude Opus 5.5 (Anthropic). Selective review and corrections: GPT-6 Astra (OpenAI). Editor: Hugo Campos. Acceptance of corrections is not product verification.
Checklist rules · Independent review and limitations · Later evidence follow-up
Earlier report
Fetch My Epic Token: CAIHL reassessment
September 8, 2026. Published AI-assisted draft.
Fetch My Epic Token enables patient-directed record retrieval. Its 88 describes infrastructure, not clinical AI output. Local scripts and hosted helpers have different data-processing boundaries, and broad no-collection wording does not explain all necessary transient processing.
Scope and agency posture
Patient-directed record retrieval and AI-enabling infrastructure. The core token utility does not itself provide AI clinical judgments.
Posture: Strongly agency-expanding, with credential-handling and technical-burden caveats.
Axis: 88/100, retained provisionally. No numerical recalibration was performed.
Sources and documented findings
- The homepage describes patient login, token/ID access and FHIR exports, and explicitly offers a hosted copy of its download script. https://fetch-my-epic-token.org/
- The policy still says no personal-information collection, retention or distribution. Public script source accepts a POST token or local environment token, creates a temporary directory and sends bearer-authorized record requests. This confirms processing, not the live server's retention or logs. https://fetch-my-epic-token.org/policies.shtml and https://raw.githubusercontent.com/glmck13/Fetch-My-Epic-Token/main/getEHR_async.py
Patient authority
Inference from the documented workflow: Patients gain portable records and can choose downstream analysis. Hosted helper use delegates sensitive processing, while local script use has a different trust boundary.
Critical capacity
Editorial assessment: Access to underlying records can support checking and advocacy. Correcting source records remains a provider process, and downstream AI quality is outside this utility.
Informed control
Editorial assessment: The policy's broad no-collection wording does not explain necessary transient server processing. Source availability enables inspection but is not an audit or evidence of deployment parity.
Assessment and limits
The strong enabling contribution is grounded in data portability, not free licensing alone. A direct comparison with conversational clinical assistants would misstate scope.
Confidence: Moderate for public interface and source design.
Remaining uncertainty: Hosted code parity, logging, temporary-data deletion behavior and practical credential warnings remain unverified
Published documentation is evidence of stated conditions, not proof of actual implementation. Unknowns did not receive automatic negative points. Funding, sponsorship, and public code do not determine agency by themselves.
Review provenance
- Reviewer/model: OpenAI Codex / GPT-6.
- Method: Focused public-source reassessment using CAIHL: patient authority, critical capacity, and informed control. Existing evidence plus one focused primary-source pass and at most one targeted follow-up. No live product testing. Numerical scores remain provisional editorial placements, not a new calculation.
- Human review: Hugo Campos authorized publication of these AI-assisted draft reassessments on September 8, 2026. This does not claim comprehensive human verification of every finding.
- Earlier review: 2026-06-12. Historical assessment. Earlier claims are not automatically reverified by this publication.